Interface: EnrollVoiceRequest

@kortexya/reasoninglayer

---

@kortexya/reasoninglayer / Speech / EnrollVoiceRequest

Interface: EnrollVoiceRequest

Defined in: src/types/speech.ts:23

Body of POST /api/v1/speech/voices — enroll (or re-enroll) a cloned voice.

Properties

audio

audio: string

Defined in: src/types/speech.ts:25

Base64-encoded reference clip (~5–10 s of clean speech works best).

---

audioMime?

optional audioMime: string | null

Defined in: src/types/speech.ts:27

MIME hint for decoding the clip. Optional.

---

consent

consent: VoiceConsent

Defined in: src/types/speech.ts:29

Who authorized this clone. Required.

---

engine?

optional engine: string | null

Defined in: src/types/speech.ts:31

Engine whose conditioning to precompute. Defaults to "qwen3-tts". Optional.

---

name

name: string

Defined in: src/types/speech.ts:36

Display name. Re-enrolling the same name (case-insensitive) overwrites the same voice rather than accumulating duplicates.

---

transcript?

optional transcript: string | null

Defined in: src/types/speech.ts:41

Transcript of the clip. Supplying it enables the ICL clone path, which measurably improves speaker similarity. Optional.